SDV Engineering Series Season 1: Building an AUTOSAR Adaptive application with Agentic AI

 SDV development is breaking the traditional V-cycle.The bottleneck isn’t writing software anymore, it’s managing distributed service-oriented architectures, keeping models, code and deployed systems in sync, and shipping updates without re-running the whole verification cycle. 

This series follows one engineer building a real AUTOSAR Adaptive application, using System Composer, Simulink and Agentic AI, from architecture definition to a live Over-the-Air update on physical hardware. 

Proof of reality  

Based on real Model-Based Design workflows used in automotive software development with System Composer, Simulink and AUTOSAR Adaptive architectures. 

What this series covers

A full SDV development lifecycle, built around one example: a Light Control System. 

Chapter 1: The basics – setting up your environment and creating Service-Oriented Architectures 

Set up the Agentic AI environment and use the MATLAB and Simulink Agentic Toolkits to create a Service-Oriented Architecture for a Light Control System. 

• Define interfaces and architecture from existing requirements 

• Use Agentic AI to accelerate architecture and Simulink implementation 

• Create a traceable workflow from requirements to design 

Chapter 2: Verification & deployment – test your design and deploy it to real hardware 

Verify the design against requirements and deploy the AUTOSAR Adaptive application to real hardware. 

• Create test cases independently of the design process using an Agentic AI workflow 

• Verify the design against requirements using a workflow compliant with functional safety standards 

• Deploy the application to a Raspberry Pi using Embedded Coder Support Package for Real-Time Linux 

By the end of this chapter, you have a working application running on real hardware. 

Chapter 3: Over-the-air updates (OTA) – maintain your software after deployment 

Modify the deployed application and deliver the updated version to physical hardware using an over-the-air updates. 

• Update the architecture and rebuild the application 

• Configure the release server and deployment workflow 

• Deliver the updated application to hardware via OTA 

Complete the end-to-end journey from requirements and AI-assisted architecture to verification, deployment, and OTA updates. 

Why this is different

No conceptual overview, no slideware. Every chapter is built on an actual toolchain run: System Composer architecture, Simulink models, generated code, deployed hardware, OTA rollout. You see where requirements become architecture, where architecture becomes executable models, and where Agentic AI shortens each step, with the trade-offs included. 

Who this is for

• Automotive software engineers 

• System and software architects 

• ADAS and SDV development teams 

• Embedded software engineers 

• MATLAB and Simulink users working in Model-Based Design 
• Engineering leaders driving software transformation programs 

What you get

• A complete AUTOSAR Adaptive workflow, chapter by chapter 

• Real System Composer and Simulink artifacts, not summaries 

• Practical Agentic AI usage patterns you can apply to your own projects 
• A working reference implementation to benchmark your own SDV development process against 

Featured products  

All products mentioned are developed by MathWorks®:  

• MATLAB® – Algorithm development, data analysis, and automation  

• Simulink® – Model-Based Design and system-level simulation  

• SciEngineer Professional Services – SDV consulting, workflow design, and toolchain integration  

 Learn more 

• Expert Insights with Bence Nagy: Software-defined Vehicles: The Platform Shift Transforming the Automotive Sector 

• Virtual ECUs and SIL Simulation: Accelerating Software-defined Vehicle Development with MATLAB & Simulink  

• The Road to Software-defined Vehicles: From Vision to Implementation  

• Software-defined Vehicles: The New Era of the Automotive Industry